Practice: Graphics with ggplot2 (part 2)

Stat 133

Learning Objectives

• Get started with "ggplot2"
• Produce basic plots with ggplot()
• Gain familiarity with the aes() function
• Learn about the various geoms, or geometric objects, and recognize them
• Understand why and how to facet
• Try out different plot themes

1 First contact with ggplot()

In this module you will learn how to create graphics with "ggplot2" which is part of the "tidyverse" ecosystem of packages.

library(tidyverse)

1.1 Data mpg

For illustration purposes we are going to use the mpg data which is one of the data sets in "ggplot2":

mpg

# A tibble: 234 × 11
   manufacturer model      displ  year   cyl trans drv     cty   hwy fl    class
   <chr>        <chr>      <dbl> <int> <int> <chr> <chr> <int> <int> <chr> <chr>
 1 audi         a4           1.8  1999     4 auto… f        18    29 p     comp…
 2 audi         a4           1.8  1999     4 manu… f        21    29 p     comp…
 3 audi         a4           2    2008     4 manu… f        20    31 p     comp…
 4 audi         a4           2    2008     4 auto… f        21    30 p     comp…
 5 audi         a4           2.8  1999     6 auto… f        16    26 p     comp…
 6 audi         a4           2.8  1999     6 manu… f        18    26 p     comp…
 7 audi         a4           3.1  2008     6 auto… f        18    27 p     comp…
 8 audi         a4 quattro   1.8  1999     4 manu… 4        18    26 p     comp…
 9 audi         a4 quattro   1.8  1999     4 auto… 4        16    25 p     comp…
10 audi         a4 quattro   2    2008     4 manu… 4        20    28 p     comp…
# ℹ 224 more rows

---

2 Example: Scatterplots

Let’s start with a scatter plot to visualize the relationship between engine displacement displ, and highway miles per gallon hwy.

ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point()

Recap:

• ggplot() creates an object of class "ggplot"

• the main input for ggplot() is data which must be a data frame

• then we use the "+" operator to add a layer

• the geometric object (geom) are points: geom_point()

• aes() is used to specify the x and y coordinates, by taking columns displ and hwy from the data frame

The same scatter plot can also be created with this alternative use of ggplot()

# scatterplot (option 2)
ggplot(data = mpg) +
  geom_point(aes(x = displ, y = hwy))

---

3 Using aes()

Does anything happen if you don’t name the arguments to aes(), i.e. you type in aes(displ, hwy)? What if you type in aes(hwy, displ)?

# Typing aes(hwy, displ) gives you another scatter plot
# in which 'hwy' goes to the x-axis, and 'displ' to y-axis
ggplot(data = mpg, aes(hwy, displ)) +
  geom_point()

Let’s restrict the data set to just cars manufactured by Audi.

audi = filter(mpg, manufacturer == 'audi')

ggplot(data = audi, aes(x = displ, y = hwy)) +
  geom_point()

3.1 Using geom_text()

Let’s label each point using model by adding a geom_text() layer, and mapping this argument with aes():

ggplot(data = audi, aes(hwy, displ)) +
  geom_point() +
  geom_text(aes(label = model))

1. The model names overlap with the points. Modify your code above by using the nudge_y argument in geom_text(). Does it go inside or outside of aes()? Now, replace geom_text() with geom_label(). What difference do you notice? Did you have to modify the arguments to aes() at all?

# argument nudge_y goes outside aes()
ggplot(data = audi, aes(hwy, displ)) +
  geom_point() +
  geom_text(aes(label = model), nudge_y = 0.1)

2. Next, cut and paste the aes(x = displ, y = hwy) from the argument of ggplot() to the argument of geom_point(). Do you run into an error? What if you copy the x and y arguments over to the aes() function in geom_text()?

# specifying a local mapping just for geom_point results in an error
ggplot(data = audi) +
  geom_point(aes(hwy, displ)) +
  geom_text(aes(label = model), nudge_y = 0.1)

---

4 Adding color

Let’s go back to the full data set.

ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point()

1. First, make all the points blue. Should you use aes()?

# to make all points blue don't use aes()
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point(color = "blue")

2. Next, color the points by class. Should you use aes() this time?

# to color points by 'class' you should use aes()
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point(aes(color = class))

3. Try coloring the points off of other variables. Which variables display a reasonable amount of colors? Which ones display far too many?

# you can color points by 'drv' (the type of drive train)
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point(aes(color = drv))

# coloring points by 'manufacturer' gives too many colors
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point(aes(color = manufacturer))

4. Also try modifying the size of points, both inside and outside of aes(). For which variables does it make sense to map them to size? For which ones does it not make sense?

# mapping 'cty' to size
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point(aes(color = class, size = cty))

# setting size to 3
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point(aes(color = class), size = 3)

---

5 Adding smoothers

Let’s fit a line to our scatter plot. To be more specific, let’s fit a linear model (i.e. least squares regression line) of hwy onto displ:

ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  geom_smooth(method = 'lm')

`geom_smooth()` using formula = 'y ~ x'

geom_smooth() with the argument method = 'lm' plots a least squares regression line for highway mileage on engine displacement. The translucent gray band is a confidence interval for the predictions of mileage.

1. Modify the code above by adding a vertical line at \(x=4\) using geom_vline(). Does it require aes()?

# adding a vertical line does not require aes()
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  geom_smooth(method = 'lm') +
  geom_vline(xintercept = 4)

2. Now try adding a vertical line at the mean of displ. Does it require aes() this time?

# adding a vertical line at the mean of 'displ'
# does require aes()
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  geom_smooth(method = 'lm') +
  geom_vline(aes(xintercept = mean(displ)))

3. Do the same with a horizontal line at the mean of hwy. Play around with color and size. Should those arguments go inside or outside of aes()?

# adding a horizontal line at the mean of 'hwy'
# does require aes()
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  geom_smooth(method = 'lm') +
  geom_hline(aes(yintercept = mean(hwy)))

---

6 Using facets

Let’s return to the basic scatter plot again.

ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point()

There are only two unique values for year, 1999 and 2008. Let’s compare the relationship between hwy and displ, distinguishing by years using facet_wrap().

ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  facet_wrap(~ year)

1. What happens when you facet (essentially, compare) using a different variable? Modify the code above and try.

# your code here
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  facet_wrap(~ cyl)

2. facet_grid() works slightly differently from facet_wrap. The latter takes in only one variable, which always goes behind the ~, and it ‘wraps’ the plots left to right, top to bottom. facet_grid() allows you to facet into just rows or just columns.

# facet into rows
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  facet_grid(. ~ year)

# facet into columns
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  facet_grid(year ~ .)

The . is a placeholder for a variable. Modify either code chunk above by replacing the . with another variable, such as cyl. How does the display change?

6.1 More facets

Finally, let’s study just the distribution of highway mileage.

ggplot(data = mpg) +
  geom_histogram(aes(x = hwy)) +
  facet_wrap(~ class)

`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

Instead of using a histogram to study the distribution, let’s try a boxplot instead.

ggplot(data = mpg) +
  geom_boxplot(aes(x = hwy))

Notice that geom_histogram() and geom_boxplot() required only an x aesthetic. What happens if you replace x with y?

1. Facet again by modifying the code above, this time using any variable of your choice.

# facets by 'drv' (the type of drive train)
ggplot(data = mpg) +
  geom_boxplot(aes(x = displ)) +
  facet_wrap(~ drv)

---

7 Using Themes

Graphics produced with ggplot() use a default theme for things such as the color of the background, the grid lines (auxiliary horizontal and vertical lines), the tick marks, position of a legend, etc.

Interestingly, you can change the appearance of a graphic by using other themes. Here’s one example with theme_bw() that uses a Black-White theme layer to the original scatter plot:

ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  theme_bw()

Look at more theme options in the "ggplot2" cheatsheet or check the help() documentation of theme functions ?theme_bw, and try at least 2 more themes:

# minimal theme
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  theme_minimal()

# classic theme
ggplot(data = mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  theme_classic()